RESEARCH ARTICLE

Climate Tolerances and Habitat Requirements Jointly Shape the Elevational Distribution of the American Pika (Ochotona princeps), with Implications for Climate Change Effects Leah H. Yandow1,2*, Anna D. Chalfoun3, Daniel F. Doak1,4 1 Department of Zoology and Physiology, University of Wyoming, 1000 East University Avenue, Laramie, Wyoming, 82071, United States of America, 2 Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology (3166), University of Wyoming, 1000 East University Avenue, Laramie, Wyoming, 82071, United States of America, 3 U.S. Geological Survey Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology (3166), University of Wyoming, 1000 East University Avenue, Laramie, Wyoming, 82071, United States of America, 4 Environmental Studies Program, University of Colorado Boulder, 1201 17th St., 397 UCB, Boulder, Colorado, 80309, United States of America OPEN ACCESS Citation: Yandow LH, Chalfoun AD, Doak DF (2015) Climate Tolerances and Habitat Requirements Jointly Shape the Elevational Distribution of the American Pika (Ochotona princeps), with Implications for Climate Change Effects. PLoS ONE 10(8): e0131082. doi:10.1371/journal.pone.0131082 Editor: Christopher A. Lepczyk, University of Hawaii at Manoa, UNITED STATES Received: May 8, 2014 Accepted: May 28, 2015 Published: August 5, 2015 Copyright: This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication. Data Availability Statement: Data will be available upon acceptance of the manuscript. We will be using Dryad repository as recommended by PLOS ONE for ecological publications. Funding: This work was funded by the following: Wyoming Governor’s Big Game License Coalition (two years of funding) to ADC (http://www. wyomingwildlifefoundation.org/), Doak Startup from CU Boulder to DFD (http://www.colorado.edu/envs/ people/dan-doak), and American Society of Mammalogists Grants In Aid to LHY (http://www. mammalsociety.org/grants). The funders had no role

* [email protected]

Abstract Some of the most compelling examples of ecological responses to climate change are elevational range shifts of individual species, which have been observed throughout the world. A growing body of evidence, however, suggests substantial mediation of simple range shifts due to climate change by other limiting factors. Understanding limiting factors for a species within different contexts, therefore, is critical for predicting responses to climate change. The American pika (Ochotona princeps) is an ideal species for investigating distributions in relation to climate because of their unusual and well-understood natural history as well as observed shifts to higher elevation in parts of their range. We tested three hypotheses for the climatic or habitat characteristics that may limit pika presence and abundance: summer heat, winter snowpack, and forage availability. We performed these tests using an index of pika abundance gathered in a region where environmental influences on pika distribution have not been well-characterized. We estimated relative pika abundance via scat surveys and quantified climatic and habitat characteristics across two North-Central Rocky Mountain Ranges, the Wind River and Bighorn ranges in Wyoming, USA. Pika scat density was highest at mid-elevations and increased linearly with forage availability in both ranges. Scat density also increased with temperatures conducive to forage plant growth, and showed a unimodal relationship with the number of days below -5°C, which is modulated by insulating snowpack. Our results provide support for both the forage availability and winter snowpack hypotheses. Especially in montane systems, considering the context-dependent nature of climate effects across regions and elevations as well as interactions between climatic and other critical habitat characteristics, will be essential for predicting future species distributions.

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

1 / 21

Climate and Habitat Shape the Distribution of the American Pika

in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist.

Introduction Though climate change is now exceptionally well supported, the continued and future ecological effects of climate change for a wide variety of species are less certain. The most compelling evidence for ecological effects of climate change comes from shifts in species distributions [1– 3]. Elevation is frequently used as a surrogate for climate in predicting shifts in range limits, and many researchers have predicted that species will migrate upslope with ongoing climate change, because temperature, and in some cases precipitation, are strongly related to elevation [4, 5]. However, syntheses suggest that many species do not actually fit the general expectations for range shifts [6]. Effects of complex topography and other co-varying habitat characteristics in mountain systems are likely to complicate the relationship between elevation and climatic conditions, and species’ responses to shifting temperatures. Thus, more accurately understanding how and why species are limited by climate and other factors in mountain habitats is an important step in assessing the ecological effects of climate change. In spite of this complexity, the multitude of alpine species that are responding to climate change demonstrates that alpine environments are some of the most susceptible habitats to changing climate [7–10]. The American pika (Ochotona princeps) is often considered one of the most sensitive alpine species due to its unusual natural history, and has been touted as an important indicator of changes in alpine habitats. Pikas are herbivorous lagomorphs that do not migrate or hibernate and maintain a high metabolism through the summer, rendering them vulnerable to the full suite of seasonal climate stressors and especially to temperature extremes [11, 12]. Within habitats of suitable climate conditions, pikas occupy specific substrates that provide refuge from predators and buffer them from thermal changes. These habitats vary across their range and include mine tailings, rocky hills, lava beds and, in the Rocky Mountains, primarily high elevation talus slopes. Since pikas are limited by environmental conditions year-round and highly associated with specific habitats [13], there is strong potential for climate, in combination with other factors, to limit their distribution. The vulnerability of the American pika to extreme high temperatures has been recognized for decades [12]. However, more recently, a focus on climate change impacts on ecological systems has prompted new research on the severity and ubiquity of these limitations. Pika populations in the Great Basin have shown patterns of extinction associated with high summer temperatures, acute cold stress, forb cover and summer precipitation [11, 14–16]. In contrast, Southern Rocky Mountain populations have shown little or no evidence of decline, but rather show local extirpations associated with consistently dry sites [17]. In other parts of the pika’s range, stable populations have been the norm, including at some exceptionally low elevation sites and across a variety of substrates [18–21]. These patterns suggest context-dependent responses to aspects of climate within a single species across its range. The American pika is therefore an ideal focal species for the investigation of geographic variation in the relative influence of climatic and habitat variables across elevation, to improve understanding of contextdependent effects of climate change. The most dramatic changes in pika distribution have been documented on their range periphery, in places that are relatively hot, dry, and often at low elevations. What remains unclear is how climatic characteristics affect pikas in more northern, mesic habitats such as within the North-Central Rocky Mountains, where pikas primarily inhabit alpine talus slopes. Hot summer temperatures may be less of a concern at these latitudes, but with increasing yearround temperature and precipitation variability, ambient climatic conditions may still limit pika populations. In particular, in high-elevation montane environments, persistent insulating snow cover is likely to be critical for over-winter survival. A combination of higher frequency of melt-freeze cycles and a lack of insulating snow may lead to ruined haypiles (drying food

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

2 / 21

Climate and Habitat Shape the Distribution of the American Pika

needed for overwinter survival) and/or acute cold stress [11]. Pikas at more northern latitudes, therefore, may not have the same climate and habitat limitations as those regions in which studies have shown higher occupancy at high elevation and on northeasterly slopes [11, 18, 21]. Moreover, few studies have evaluated the relative abundance of pikas in relation to potential limiting factors, instead using presence/absence data to index local populations (but see 14, 22). Estimates of abundance may provide more nuanced inference about habitat suitability than these occupancy indices. We analyzed climate-related habitat features (i.e., elevation, aspect, elevation difference to summit) and other habitat variables thought to be important for pika persistence (i.e., talus depth and forage availability) and their joint effects on pika relative abundance as indexed by scat density. We also tested for effects of several derived climate variables on scat density and investigated which of these climate variables are indexed by elevation at our sites. Our primary objectives were to 1) evaluate which habitat features may be most limiting to pikas in the northern part of their range, including both features thought to index some aspect of climate and other local habitat features identified as potentially important for persistence, 2) test the generality of these patterns across two similar yet geographically distinct mountain ranges in the North-Central Rocky Mountains, and 3) use derived climate variables from temperature sensors to investigate which aspects of climate most co-vary with elevation in this region. We first looked for patterns in scat density with elevation and then tested three non-mutually-exclusive hypotheses concerning the factors limiting pika abundance and distribution, each based on previous research and the natural history of the American pika. The summer heat hypothesis suggests that high summer temperatures limit pika numbers via acute and chronic heat stress and reduced foraging and caching time [11, 12, 23]. Under this hypothesis we expected low pika abundance at characteristically warmer sites based on habitat features, such as at low elevation and southern aspects. The winter snowpack hypothesis emphasizes the importance of an insulating snowpack during winter whereby variable snowpack could lead to cold stress or death either by exposure to extreme cold, higher predation rates, and/or starvation by damaged haypiles [11]. If winter snowpack was influencing pika abundance at our sites, we predicted lower pika abundance at sites prone to variable snowpack conditions such as near the summit, on wind-scoured slopes, or particularly low elevation sites where repeated melt-freeze cycles are more likely to occur. Finally, the forage availability hypothesis suggests that food availability has a strong limiting effect on pika numbers and distribution [24–26]. If forage availability was a primary limiting factor on pikas, we expected a positive association between pikas and both forage availability and climatic variables that create favorable plant growth conditions during summer. While environmental conditions may also modify haying behavior, we reasoned that the overall amount of forage available should still influence the numbers of pikas within an area. Overall, our aim was to evaluate the support for these three hypotheses and thereby investigate which factors may be most limiting to pika relative abundance in our region.

Materials and Methods Approach We selected climatic and habitat variables related to the predictions of our three hypotheses (Table 1) and assessed pika relative abundance within two distinct mountain ranges in Wyoming, USA, using scat density as an index. We tested general linear models including combinations of elevation, aspect, elevation difference to summit, talus depth, and two metrics of forage availability as potential predictive factors (Table 1), and judged the support for each using information criteria (AICc). We then fit a series of models that included factors other than elevation in the best model of the first analysis but replaced elevation with climatic variables (days above

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

3 / 21

Climate and Habitat Shape the Distribution of the American Pika

Table 1. Habitat variables (top) and derived climate variables (bottom) used in general linear models and model selection analyses of relative pika (Ochotona princeps) scat density in the Wind River (2010) and Bighorn (2011) mountain ranges, Wyoming, USA. Variable

Relationship to Climate

Elevation

Inversely related to temperature

Relevant Hypotheses Summer heat; winter snowpack

Aspect

South and west are warmer aspects

Summer heat; winter snowpack

Elevation difference to summit

Index of wind and melt/freeze exposure

Summer heat; winter snowpack

Patch forage

N/A

Forage availability

Perimeter forage

N/A

Forage availability

Talus depth

N/A

N/A

Variable

Description

Relevant Hypothesis

days above 15°C

Total of days temperature  15

Summer heat

winter mean

Average temperature of days < 0

Snowpack

days below 0°C

Total # of days temperature < 0

Snowpack

days below -5°C

Total # of days temperature < -5

Snowpack

days below -10°C

Total # of days temperature < -10

Snowpack

days above 10°C

Total # of days temperature < 10

Forage availability

total degree days

Total # days above 0 x average temperature

Forage availability

length of growing season

Total # of days above 0

Forage availability

summer mean

Average temperature of days > 0

Forage availability

N/A = variables included in models as a necessary component of pika habitat but not considered a potential proxy for climate. doi:10.1371/journal.pone.0131082.t001

15°C, winter mean temperature, days below 0°C, days below -5°C, days below -10°C, days above 10°C, total degree days, length of growing season, and mean summer temperature) derived from the local temperature data from one of the mountain ranges in a (Table 1). Finally, we tested a full model suite of climatic variables and their ability to predict pika scat abundance.

Study sites Our sites were located within the Wind River (hereafter Winds) and Bighorn mountain ranges, which are separated by a 200 km-wide basin of sagebrush steppe. We chose the Winds for its variation in topography and climate and abundant alpine habitat. Using aerial imagery, we selected between 30 and 40 potential survey sites in each of four quadrants: northeast, southeast, southwest, and northwest. At least eight of those sites were in each of three elevation classes: < 3,000 m; 3,000–3,500 m; and > 3,500 m. Similarly, we stratified across northeast, southeast, southwest, and northwest aspects, and within each elevation group assigned at least two sites for the four aspects. We used the intercardinal directions because prevailing weather patterns in the Winds are primarily northwesterly and southeasterly. To ensure that potential sites covered a broad range of available climatic conditions, we visually assessed precipitation and maximum and minimum temperatures from Parameter-elevation Regression on Independent Slopes Model (PRISM) data for January and July of each year from 2000–2009, choosing potential sites that covered a broad range of pixel values for each of the climate variables. Similarly, we used National Agricultural Imagery Program (NAIP) aerial imagery from July 2009 to evaluate variation in forage availability and select sites with a wide range of number of red pixels, indicative of vegetation, within and around the potential survey area. We also chose sites with variation in microclimate as indicated by snow presence in the imagery. In our list of potential sites, we included those with both lingering snowfields and early snowmelt, as well as sites with evident patches of vegetation and those with little to no apparent vegetation. We ultimately selected 43 survey sites (i.e., talus patches; elevation 2540–3926 m) in 2010, based on sampling across the broadest range of attributes available and accessibility. We considered a

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

4 / 21

Climate and Habitat Shape the Distribution of the American Pika

patch of talus or potential survey site to be a rock field primarily consisting of rocks > 0.5 m on their longest axis that were often interspersed with small alpine meadows or other vegetation. We defined the size and boundaries of each survey site based on local features such as a cliff band, a marked change in aspect, or conspicuous line of vegetation such as trees or a meadow. To ensure that we sampled potentially suitable sites, each site used in the analysis had to include at least 75% talus, as ground-truthed in the field. We resurveyed 9 of the 43 sites in 2011 to test for potential year effects. Resurveyed sites included samples from each of the four regions of the range, as well as sites at low, mid, or high elevations, and spanned all aspects. In 2011, we conducted surveys in our second focal mountain range, the Bighorns. Similar in geology, climate, and wildlife communities, the Bighorns served as an appropriate comparison area to test the generality of patterns observed in the Winds. We chose 60 potential sites in the Bighorns using the same methodology of stratifying survey sites across habitat and climate attributes. Based on field logistics and accessibility, we ultimately surveyed 40 of those sites (2158–3897 m) from 25 June -13 August 2011.

Ethics Statement All data collection for our study was purely observational and did not involve a threatened or endangered species and therefore the only permits obtained for this research were for access on the Wind River Reservation for 2 of the 83 survey sites. We obtained permission to access those sites through the Tribal Fish and Game. All other sites in the Winds were located within the Shoshone and Bridger-Teton National Forests (between 12T 593644mE, 4804674mN and 668445mE, 4722213m N; NAD 83) and therefore did not require permission for access. Sites in the Bighorns were all on public land in Bighorn National Forest (between 13T 322689mE, 4927468mN and 344272mE, 4890903mN; NAD 83) and therefore did not require permission for access.

Pika Relative Abundance Pikas are typically studied using occupancy methods due to logistical constraints [14]. Presence/absence data alone, however, yield limited ecological inference. We therefore used pika scat density as an index of relative abundance, as a compromise between occupancy and density estimation. Scat counts have been one of the most widely used methods for quantifying the relative abundance of mammalian species [27–31], and occupancy in lagomorphs [17, 31–34] including pikas [22], because they provide a more stable record of presence and habitat use than do visual observations. Pika scat is conspicuous and often persists for multiple years [35]. Additionally, large sample sizes of spatially-independent sites spanning a diversity of climatic and other habitat variation were essential for testing our hypotheses. For broad, multi-site studies like ours, scat surveys are feasible to conduct across a large number of sites within a reasonable amount of time. We acknowledge several caveats related to our use of scat density for estimating pika numbers. Scat degrades over time, and possibly at an inconsistent rate across sites depending on moisture levels. There is also some error associated with counting scat, as inevitably some piles will not be located. However, we assume that environmental conditions are similar enough across sites and that decomposition rates should also have been similar. We also very carefully attempted to locate all scat piles, and assumed that scat that was unaccounted for was evenly distributed across sites. Direct abundance estimates are indeed ideal for more accurate delineation of proximate abundance and temporal comparisons of abundance. Their usefulness for characterizing factors influencing the distribution of a species known for its metapopulation dynamics and weather-dependent behavior, however, is limited. Methods for assessing pika

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

5 / 21

Climate and Habitat Shape the Distribution of the American Pika

abundance that rely on aural and/or visual detections, for example, are subject to the vagaries of ambient weather conditions and are time-intensive [14], rendering them impractical for studies necessitating broad-scale spatial replication in remote and widely dispersed survey sites. Similarly, using only very recent indicators of abundance, such as fresh sign, can introduce confounding effects of inter-annual variation in numbers that are, from the point of view of our questions, statistical noise. We therefore concluded that scat density was the best metric for indexing general abundance across several sites within a single year, as a time-averaged signal of local habitat suitability. We also assumed that pika abundance across years, as indexed by scat density, was a reliable indicator of habitat suitability, a reasonable assumption given pika life history. Pikas are sitefaithful and have small home ranges. After dispersal, they occupy a territory for their lifetime without seasonal movement and are therefore susceptible to whatever ambient abiotic and biotic conditions exist at that particular site. With very little movement of individuals post-dispersal we expected that the density of scat, and therefore pikas, in a patch should be a good indicator of how well that patch consistently supported pikas during recent years. Therefore, using scat as an estimate of moderate-term abundance allowed us to infer the suitability of an area for pika persistence, an important consideration for a species that is well-known to show transient extirpations and recolonizations at local sites [36–38]. The method also enabled us to avoid the potential confounding effects of ephemeral weather influences that can skew other indices such as aural or visual pika detections. Even though pikas are typically conspicuous, weather and season strongly influence detections of individuals within a single day or across a season. We also considered using haypiles or only fresh scat as an indicator of pikas. However, as proxies for abundance, these variables are also subject to temporal inconsistency. Haypiles are usually difficult to detect before August and fresh scat is not simple to score with high repeatability (Yandow, pers. obs.), and indexes only current year abundance, thereby ignoring the inter-annual fluctuations in abundance. Haypiles were moderately correlated with pika scat in the Winds in 2010 (r = 0.29, N = 43, P = 0.05) and strongly correlated with scat in the Bighorns in 2011 (r = 0.78, N = 40, P < 0.01), providing additional evidence that scat was a reasonable predictor of pika relative abundance. We sampled scat at each survey site along parallel line transects [33, 34, 39]. We established the starting point for each transect in one of three ways, as dictated by individual field situations: 1) at a talus/vegetation interface; 2) at a distinct natural feature of the landscape (rock outcropping, cliff edge, etc.); or 3) where the aspect for the particular site changed. Size and shape of the talus patch determined the number and length of transects [40]. There were 1–5 transects per site that ranged in length from 53 to 254 m. In cases where sites were established within an entire hillside or ridgeline of continuous talus habitat, we sampled three transects 210 m in length. While the size of pika territories can vary, we assumed them to average ~30 m in diameter [12]. Our approach, therefore, allowed for survey of ~2–9 possible pika territories per transect but ultimately accounted for scat within 2 m of the transect line to give us an estimate of scat density. This variable sampling effort, dependent on site size, was similar to that employed by another recent paper on pika distribution [22]. We decreased the 210 m standard to two 150 m transects for the Bighorn sites in 2011 to increase survey efficiency. We considered this reasonable because we maintained consistent transect placement and sampling methods. Parallel transects were 60 m from one another on the slope except for sites smaller than 8 pika home ranges. In such cases, we placed parallel transects 30 m from one another to allow for approximately 1 pika home range between surveyors (6 sites of 83). Surveyors slowly moved across talus slopes perpendicular to the fall line searching along a given transect for all scat within 2 m on either side, including within crevices and rock interstices. We counted pellets singly up to piles of 25, and due to the difficulty of reliably

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

6 / 21

Climate and Habitat Shape the Distribution of the American Pika

distinguishing between old and new scat, did not record these separately. If a pile had more than 25 pellets, we simply recorded it as 25. Another study has used this number as a threshold to determine site occupancy [18]. Pika scat tends to be clumped, which made it suitable to record pellets within 1 m of a pile as part of the same pile. We calculated density of scat by dividing by the number of meters surveyed for all transects at a given site. In a subset of resampled sites, scat counts were significantly higher in 2011 (2010: X = 0.28, SD = 0.20; 2011: X = 0.57, SD = 0.31; t(9) = -6.015, P < 0.001). However, scat counts by site across years were highly correlated (r = 0.93, N = 7, P < 0.001).

Surveyor Bias Four observers participated in our surveys in each year (2010 and 2011), with two participating both years, for a total of six unique observers. To minimize potential observer bias, each year observers were randomly assigned across varying site strata including elevation, aspect and slope. To test for potential bias, in 2011 each of the four researchers surveyed the same extra 60-m “test transects” (n = 10). We used these data in a one-way ANOVA to test for observer differences. There was no difference between observers (F2, 27 = 0.69, P > 0.05), and we assumed potential observer differences were similar in 2010.

Forage Availability We measured forage availability via two metrics: the abundance of foraging habitat within a patch (“patch forage”) and the proportion of the perimeter of the talus patch that was foraging habitat (“perimeter forage”). At each site we surveyed forage availability from the middle of the second transect or from a haypile within 12 m of that point when available and used a modified point-intercept method [16]. In a few cases, the middle of the second transect was considered unrepresentative of the site because it landed either on a rock outcropping or in the center of the only meadow in the site. In such cases, we moved the survey point down slope 30 m. At each survey point for patch forage, we established two perpendicular 50-m transects along which vegetation cover types were recorded at 1-m intervals (n = 100 points per survey). The number of points out of 100 that touched potential forage was our estimate of percent forage availability. We included grasses, forbs, shrubs, trees, cushion plants, mosses, and ground lichens as available forage because 1) pikas are generalist foragers [41] and 2) we observed all of these forage types in pika haypiles. We also estimated forage availability around the edge of each site to account for sites that are primarily rock within the talus patch but have potential foraging areas around the perimeter [42]. We designated each meter along the scat survey line as available forage if there was a distinct edge of meadow within 15 m (visual estimate of greater than 75% vegetation cover). We divided the number of meters characterized as available forage by the total meters surveyed to calculate the proportion of perimeter forage availability for each site.

Climatic Variables We used Thermochron iButton temperature sensors to measure ambient temperatures at sites in the Winds during 19 September 2010 through 17 August 2011 (model DS1921G; temperature ranges -40°C to 85°C). We set all sensors to track temperature every four hours (02:00, 06:00, 10:00, 14:00, 18:00, and 22:00 hours each day). We deployed a sensor at the approximate center of each survey site, near a haypile when available, to record ambient temperatures that a pika at that site would likely experience. We sealed each sensor in a 0.5-ounce clear plastic case with approximately one half of a teaspoon of anhydrous calcium chloride desiccant (DampRid). Each sensor was affixed to a rock using clear polyvinyl tape and heavy-duty weed whacker line. The

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

7 / 21

Climate and Habitat Shape the Distribution of the American Pika

sensor was deployed approximately 0.75 m below the talus surface. Upon retrieval, several loggers were either missing, moved, or exposed to direct sunlight. Out of 43 sites, we retrieved 27 subsurface loggers that tracked temperature and were not exposed to direct sunlight. We derived nine climate variables from the temperature sensor data related to our three hypotheses: number of days above 15°C, number of days temperature reaching 3600 m, (n = 4). The curve represents the predicted relationship from the top model of the temperature sensor analysis (adj. r2 = 0.29). doi:10.1371/journal.pone.0131082.g004

Our results supported the forage availability hypothesis in two ways. First, the best-supported model included a positive effect of forage availability on pika scat density. Forage availability within talus patches and around patch perimeters were among the best predictive variables for the Wind River and Bighorn ranges, respectively. While these metrics indexed different aspects of food availability, both are indicative of the same general effect. Since food availability commonly influences species’ abundance, the positive linear relationship of forage and scat density was not surprising. However, along with elevation, our results suggest that forage availability plays a particularly important role in this system. Some alpine areas are comprised of talus/meadow mosaics and provide a heterogeneous landscape with strong microclimatic variation, which promotes species richness of plants, insects, and mammals [51–53]. Although pikas can range up to hundreds of meters in search of forage [54], such diverse and patchy landscapes allow pikas to collect hay more locally, which is more energetically efficient. Abundant and nearby forage enables pikas to cache as much vegetation as possible during the short alpine growing season while still being able to defend collected hay. Pikas may also consume uncollected forage within their territory if accessible under the snow during winter, which can improve the odds of over-winter survival. Another typical kind of talus patch consists of large lobes of pure talus with very little forage available within the talus matrix. Often, these patches are characterized by a distinct talusmeadow edge and high density of pikas near the interface, which provides an abundant and diverse food source [12, 42]. Such nearby meadows are often local hot spots for biodiversity [55] and may allow pikas to selectively collect forage for both summer and winter diets [56].

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

14 / 21

Climate and Habitat Shape the Distribution of the American Pika

Fig 5. Correlations between four local climate measures and elevation. Climate variables included from top to bottom: number of days above 10°C, adj. r2 = 0.34; mean summer temperature, adj. r2 = 0.28; total degree days (°C*days), adj. r2 = 0.33; and mean winter temperature, adj. r2 = 0.09. All data were obtained via ibutton sensors deployed at 27 sites in the Wind River Range, Wyoming, USA for one year starting in August 2010. doi:10.1371/journal.pone.0131082.g005

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

15 / 21

Climate and Habitat Shape the Distribution of the American Pika

Fig 6. Correlations between American pika scat density and four local climate measures. Climate variables from top to bottom: number of days above 10°C, adj. r2 = 0.06; mean summer temperature, adj. r2 = 0.02; total degree days, adj. r2 = 0.04; and mean winter temperature, adj. r2 = 0.01; were obtained via ibutton sensors in the Wind River Range, Wyoming, USA deployed during August, 2010 –August 2011. doi:10.1371/journal.pone.0131082.g006

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

16 / 21

Climate and Habitat Shape the Distribution of the American Pika

Forage availability in both types of suitable talus patches, one with a distinct talus-meadow interface and the other with patchy meadows among the talus field, was an important predictor of scat density. The forage availability hypothesis was also supported by the climatic variables we tested. Climate factors indicating warmer summer conditions (mean summer temperature, days above 10°C, and total degree days) had substantial explanatory strength. All of these factors were positively associated with scat density, and these relationships are likely related to food availability. Several experimental manipulations have shown that environmental conditions such as earlier springs and warmer growing season temperatures allow for increased plant growth and reproduction in high elevation ecosystems [44,57,58]. For herbivores, plant productivity enhances body condition and potential fitness, which may scale up to influence abundance and distributions [59]. Favorable climatic conditions can also promote higher quality forage for pikas [22]. Pika scat density was higher at sites where climate factors were favorable for forage growth suggesting that limitation on pika distributions at this latitude are dependent on plant growing conditions. Temperature sensors were placed within the talus, so caution is needed in interpreting the recorded temperatures as those experienced by forage plants. We assumed, however, that these temperatures were correlated with those experienced by surrounding vegetation within the study site. Our results also supported predictions of the winter snowpack hypothesis. The number of days below -5°C was positively correlated with scat density up to about 120 days, and negatively related at higher values. Because the days below -5°C variable indicates where temperatures are low and thermal insulation is weak, the result emphasizes the importance of insulative snowpack for pikas, but only at relatively high numbers of cold days. Across all elevations, several sites showed lower scat density with low numbers of particularly cold days. This result suggests that up to a point, more cold days may be beneficial to pikas, perhaps because they indicate conditions in which a stable insulating snowpack can be established. The relationship may reach a threshold, however, beyond which there are simply too many very cold days for pikas to withstand, such as on high elevation, wind-scoured slopes. Our work highlights the context-dependent nature of climatic effects on species across elevations within and across regions. The factors limiting American pikas across the two mountain ranges in the North-Central Rocky Mountains in our study appear to be different than correlates limiting pika range limits in other regions such as the Great Basin [11, 15, 16], where combinations of acute cold, and heat in addition to other factors appear to be important drivers of persistence. We emphasize that the highest elevation sites may not be a suitable refuge from climate change at all latitudes. Pikas may already be at their upper elevation limit in parts of their range [60], which is contrary to the idea that conditions at lower elevations are the primary limit on current pika distributions. With a lack of extreme high values in our temperature data and little support for the above 15°C climate variable, there was no support for the summer heat hypothesis, suggesting that pikas at this latitude are currently not limited by hot summer temperatures. Elevation may limit pikas and other alpine species in future warmer years, but we expect that these physiological limits will be strongly modified by constraints imposed by food availability. Our results demonstrate a complex set of climatic and elevation effects that make forecasting from current elevation and climate relationships to future conditions more ambiguous. In future warmer years, additive and interactive effects of climatic conditions and alpine habitat distribution will likely influence pika range shifts. Further work investigating the influence of climatic variables on abundance (e.g., [17]) as well as specific forage requirements would add to our understanding of the complexities of this climate-species relationship, and facilitate the predictability of pika and alpine-meadow distribution and abundance.

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

17 / 21

Climate and Habitat Shape the Distribution of the American Pika

While several alpine species worldwide have responded to changing climate by moving upslope [8, 61, 62], our findings suggest that such responses by American pikas could be severely limited by the interactive effects of elevation and food availability and quality. Under current conditions there appears to be broad overlap of suitable vegetation communities and climate conditions for pikas in the North-Central Rocky Mountains, but these zones might diverge with rapid climate change. Plant species within the alpine zone are likely to respond in a variety of ways to climate change, but in general, development of highly productive alpine meadows at higher elevations will require substantial soil development, resulting in long lags between warming temperatures and establishment of suitable forage conditions for pikas [63, 64]. We speculate that a combination of rapid upward movement of climate conditions suitable for pikas, but only slow migration of suitable plant communities, is likely to create a much narrower zone of inhabitable conditions for pikas over the near term. There is ongoing conservation concern for this species that resulted in a petition for listing under the Endangered Species Act in 2010. As climate change continues and novel climatic conditions continue to emerge, implications for the American pika will likely involve combinations of physiological and/or food resource limitation. We suggest that effective pika conservation will therefore require a comprehensive and multifaceted consideration of this species’ limitations.

Supporting Information S1 File. Supporting information for model results and correlation matrices. General linear model and AICc model selection results—Wind River Range 2010 (Table A). General linear model and AICc model selection results—Bighorn Range 2011 (Table B). General linear model selection AICc results—subset of Wind River sites 2010 (Table C). Correlation matrix for habitat variables in the Wind River Range 2010 and Bighorn Range 2011 (Table D). Correlation matrix for forage variables and climate variables in the Wind River Range 2010 (Table E). (DOCX) S2 File. All habitat and scat data for each site collected in 2010 for the Wind River Range, WY (n = 43; Table A) and in 2011 for the Bighorn Range (n = 40; Table B). (DOCX) S3 File. Climate data derived from temperature-sensors data collected 19 September 2010 through 17 August 2011 for sites in the Wind River Range (n = 27). (DOCX)

Acknowledgments We would like to extend a sincere thank you to the editor and reviewers of PLOS One for their feedback and critique of our drafts, which helped improve this manuscript. Thank you to Wyoming Game & Fish Department biologists including B. Oakleaf, M. Grenier, and B. Lanka who supported this research. A. Larsen in the Wyoming Cooperative Fish and Wildlife Research Unit provided logistical support. Special thanks to the several field assistants who helped with data collection. Any use of trade or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Author Contributions Conceived and designed the experiments: LHY ADC DFD. Performed the experiments: LHY ADC DFD. Analyzed the data: LHY ADC DFD. Contributed reagents/materials/analysis tools: LHY ADC DFD. Wrote the paper: LHY ADC DFD.

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

18 / 21

Climate and Habitat Shape the Distribution of the American Pika

References 1.

Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate change impacts across natural systems. Nature 421: 37–42. PMID: 12511946

2.

Moritz C, Patton JL, Conroy CJ, Parra JL, White GC, Beissinger SR (2008) Impact of a century of climate change on small-mammal communities in Yosemite National Park, USA. Science 322: 261–264. doi: 10.1126/science.1163428 PMID: 18845755

3.

Burrows MT, Schoeman DS, Buckley LB, Moore P, Poloczanska ES, Brander KM, et al. (2011) The pace of shifting climate in marine and terrestrial ecosystems. Science 334: 652–655. doi: 10.1126/ science.1210288 PMID: 22053045

4.

Kelly AE, Goulden ML (2008) Rapid shifts in plant distribution with recent climate change. Proc Natl Acad Sci 105: 11823–11826. doi: 10.1073/pnas.0802891105 PMID: 18697941

5.

Chen I, Hill JK, Ohlemuller R, Roy DB, Thomas CD (2011) Rapid range shifts of species associated with high levels of climate warming. Science 333: 1024–1026. doi: 10.1126/science.1206432 PMID: 21852500

6.

Lenoir J, Gegout JC, Guisan A, Pascal V, Wholgemuth T, Zimmerman NE, et al. (2010) Going against the flow: potential mechanisms for unexpected downslope range shifts in a warming climate. Ecography 33: 295–303.

7.

Dullinger S, Dirnbock T, Grabherr G (2004) Modeling climate change-driven treeline shifts: relative effects of temperature increase, dispersal and invasibility. Ecology 92: 241–252.

8.

Walther GR, Post EP, Convey P, Menzel A, Parmesan C, Beebee CP, et al. (2002) Ecological responses to recent climate change. Nature 416: 389–395. PMID: 11919621

9.

Rixen C, Dawes MA, Wipf S, Hagedorn F (2012) Evidence of enhanced freezing damage in treeline plants during six years of CO2 enrichment and soil warming. Oikos 121: 1532–1543.

10.

Grabherr G, Gottfried M, Pauli H (1994) Climate effects on mountain plants. Nature 369: 448.

11.

Beever EA, Ray C, Mote PW, Wilkening JL (2010) Testing alternative models of climate-mediated extirpations. Ecol Appl 20: 164–178. PMID: 20349838

12.

Smith AT (1974) Distribution and dispersal of pikas: influences of behavior and climate. Ecology 55: 1368–1376.

13.

Morrison SF, Hik DS (2007) Demographic analysis of a declining pika Ochotona collaris population linking survival to broad-scale climate patterns via spring snowmelt patterns. J Anim Ecol 76: 899–907. PMID: 17714268

14.

Beever EA, Brussard P, Berger J (2003) Patterns of apparent extirpation among isolated populations of pikas (Ochotona princeps) in the great basin. J Mammal 84: 37–53.

15.

Beever EA, Dobrowski SZ, Long J, Mynsberge AR, Piekielek NB (2013) Understanding relationships among abundance, extirpation, and climate at ecoregional scales. Ecology 94: 1562–1571.

16.

Wilkening JL, Ray C, Beever EA, Brussard PF (2011) Modeling contemporary range retraction in Great Basin pikas (Ochotona princeps) using data on microclimate and microhabitat. Quat Int 235: 77–88.

17.

Erb LP, Ray C, Guralnick R (2011) On the generality of a climate-mediated shift in the distribution of the American pika (Ochotona princeps) Ecology 92: 1730–1735. PMID: 21939069

18.

Millar CI, Westfall RD (2010) Distribution and Climatic Relationships of the American Pika (Ochotona princeps) in the Sierra Nevada and Western Great Basin, U.S.A.; Periglacial Landforms as Refugia in Warming Climates. Arct Antarct Alp Res 42: 76–88.

19.

Beever EA, Wilkening JL, McIvor DE, Weber SS, Brussard PF (2008) American pikas (Ochotona princeps) in northwestern Nevada: a newly discovered population at a low-elevation site. West N Am Nat 68: 8–14.

20.

Simpson WG (2009) American pikas inhabit low elevation sites outside the species’ previously described bioclimatic envelope. West N Am Nat 69: 243–250.

21.

Rodhouse TJ, Beever EA, Garrett LK, Irvine KM, Jeffress MR, Munts M, Ray C (2010) Distribution of pikas in a low-elevation lava landscape: conservation implications from the range periphery. J Mammal 91: 1287–1299.

22.

Erb LP, Ray C, Guralnick R (2014) Determinants of pika population density vs. occupancy in the Southern Rocky Mountains. Ecol Appl 24: 429–435. PMID: 24834730

23.

MacArthur RA, Wang LCH (1973) Physiology of thermoregulation in the pika, Ochotona princeps. Can J Zool 51:11–16. PMID: 4696076

24.

Golian SC (1985) Population dynamics of pikas: effects of snowpack and vegetation. Thesis. University of Colorado at Boulder.

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

19 / 21

Climate and Habitat Shape the Distribution of the American Pika

25.

Smith AT (1978) Comparative demography of pikas (Ochotona): effect of spatial and temporal age-specific mortality. Ecology 59: 133–139.

26.

Morrison SF, Pelchat G, Donohue A, Hik DS (2009) Influence of food hoarding behavior on the overwinter survival of pikas in strongly seasonal environments. Oecologia 159: 107–116. doi: 10.1007/ s00442-008-1197-5 PMID: 18987896

27.

Sadliers LM, Webbon CC, Baker PJ, Harris S (2004) Methods of monitoring red foxes Vulpes vulpes and badgers Meles meles: are field signs the answer? Mamm Rev 34: 75–98.

28.

Barea-Azco JM, Virgo E, Ballesteros-Dupero E, Moleo M, Chirosa M (2007) Surveying carnivores at large spatial scales: a comparison of four broad-applied methods. Biodivers Conserv. 16: 1213–1230.

29.

Bouche P (2007) Northern Elephant Survey. Pachyderm 42: 58–66.

30.

Ross S, Munkhtsog B, Harris S (2010) Dietary composition, plasticity, and pretty selection of Pallas’s cats. J Mammal 91: 811–817.

31.

Palomares F (2001) Comparison of 3 methods to estimate rabbit abundance in a Mediterranean environment. Wildl Soc Bull 29: 578–585.

32.

Krebs CJ, Gilber BS, Boutin S, Boonstra R (1987) Estimation of snowshoe hare population density from turd transects. Can J Zool 65: 565–567.

33.

Krebs CJ, Boonstra R, Nams V, O’Donoghue M, Hodges KE, Boutin S (2001) Estimating snowshoe hare population density from pellet plots: a further evaluation. Can J Zool 79: 1–4.

34.

Murray DL, James R, Ellsworth E, Wirsing AJ, Steury TD (2002) Estimating low-density snowshoe hare populations using fecal pellet counts. Can J Zool 80: 771–781.

35.

Nichols LB (2010) Fecal pellets of American pikas (Ochotona princeps) provide a crude chronometer for dating patch occupancy. West N Am Nat 70: 500–507.

36.

Smith AT (1980) Temporal changes in insular populations of the pika (Ochotona princeps) Ecology 61: 8–13.

37.

Franken RJ, Hik DS (2004) Influence of habitat quality, patch size and connectivity on colonization and extinction dynamics of collared pikas Ochotona collaris. J Anim Ecol 73: 889–896.

38.

Fleishman E, Ray C, Sogren-Gulve P, Boggs CL, Murphy DD (2002) Assessing the Roles of Patch Quality, Area, and Isolation in Predicting Metapopulation Dynamics. Conserv Biol 16: 706–716.

39.

Mills SL, Griffin PC, Hodges KE, McKelvey K, Ruggiero L, Ulizio T (2005) Pellet count indices compared to mark-recapture estimates for evaluating snowshoe hare density. J Wildl Manage 69:1053– 1062.

40.

Heyward MW, de Tores PJ, Dillon MJ, Fox BJ, Banks PB (2005) Using fecal pellets along transects to estimate quokka (Setona brachyurus) population density. Wildl Res 32: 503–507.

41.

Dearing MD (1996) Disparate determinants of summer and winter diet selection of a generalist herbivore Ochotona princeps. Oecologia 108: 467–478.

42.

Huntly NJ, Smith AT, Ivins BL (1986) Foraging behavior of the pika (Ochotona princeps) with comparisons of grazing versus haying. J Mammal 67: 139–148.

43.

Arft AM, Walker MD, Gurevitch J, Alatalo JM, Bret-Harte MS, Dale M, et al. (1999) Responses of tundra plants to experimental warming: meta-analysis of the international tundra experiment. Ecological Monographs 69: 491–511.

44.

Walker MD, Wahren CH, Hollister RD, Henry GH, Ahlquist LE, Alatalo JM, et al. (2006) Plant community responses to experimental warming across the tundra biome. Proceedings of the National Academy of Sciences of the United States of America 105: 1342–1346.

45.

Wipf S, Stoeckli V, Bebi P (2009) Winter climate change in alpine tundra: plant responses to changes in snow depth and snowmelt timing. Climate Change 94: 105–121.

46.

McKelvey KS, Copeland JP, Shwartz MK, Littell JS, Aubury KB, Squires JR, et al. (2011) Climate change predicted to shift wolverine distributions, connectivity, and dispersal corridors. Ecol Appl21: 2882–2897.

47.

Canon J, Dominguez F, Valdes JB (2011) Vegetation responses to precipitation and temperature: a spatiotemporal analysis of ecoregions in the Colorado River Basin. Int J Remote Sens 32: 5665–5687.

48.

Brenan A, Cross PC, Higgs M, Beckmann JP, Klaver RW, Scurlock BM, et al. (2013) Inferential consequences of modeling rather than measuring snow accumulation in studies of animal ecology. Ecol Appl 22: 643–653.

49.

Carlson BZ, Munroe JS, Hegman B (2011) Distribution of alpine tundra in the Adirondack mountains of New York U.S.A. Arct Antarct Alp Res 43: 331–342.

50.

Anderson DR, Burnham KP (2002) Avoiding pitfalls when using information-theoretic methods J Wildl Manage 912–918.

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

20 / 21

Climate and Habitat Shape the Distribution of the American Pika

51.

Haslett JR (1994) Community structure and the fractal dimensions of mountain habitats. J Theor Biol 167: 407–411.

52.

Haslett JR (1997) Mountain ecology: organism response to environmental change, an introduction. Global Ecol Biogeogr 6: 3–6.

53.

Ruzicka V (1993) Stony debris ecosystems- sources of landscape diversity. Ekologia Bratislava 12: 291–298.

54.

Huntly NJ (1987) Influence of refuging consumers (pikas- Ochotona princeps) on Sub-alpine meadow vegetation. Ecology 68: 274–283.

55.

Körner C (2003) Alpine Plant Life: Functional Plant Ecology of High Mountain Ecosystems 2nd edition. Springer-Verlag Berlin Heidelberg New York pp 47–128.

56.

Dearing MD (1997) Effects of Acomastylis rossii tannins on a mammalian herbivore, the North American pika, Ochotona princeps. Oecologia 109: 122–131.

57.

Totland Ø (1997) Limitations on reproduction in alpine Ranunculus acris. Can J Bot 75: 137–144.

58.

Walker MD, Walker DA, Welker JM, Arft AM, Bardsley T, Brooks PD, et al. (1999) Long-term experimental manipulation of winter snow regime and summer temperature in arctic and alpine tundra. Hyrol Process 13: 2315–2330.

59.

MacArthur RA, Pianka ER (1966) On optimal use of a patchy environment. Am Nat 100: 603–609.

60.

Jeffress MR, Rodhouse TJ, Ray C, Wolff S, Epps CW (2013) The idiosyncrasies of place: geographic variation in the climate-distribution relationships of the American pika. Ecol Appl 23: 864–878. PMID: 23865236

61.

Luckman B, Kavanagh T (2000) Impact of Climate Fluctuations on Mountain Environments in the Canadian Rockies Ambio 29: 371–380.

62.

Benson D, Cummins M (2011) Move, adapt, or die: Lagopus leucura changes in distribution, habitat, and number at Glacier National Park, Montana. In: R. T. Watson RT, Cade TJ, Fuller M, Hunt G, Potapov E, editors. Gyrfalcons and Ptarmigan in a Changing World, Volume I. pp. 237–246.

63.

Pennington W (1986) Lags in adjustment of vegetation to climate caused by the pace of soil development. Evidence from Britain. Vegetatio 67: 05–118.

64.

Huntly NJ (1991) Herbivores and the dynamics of communities and ecosystems. Annu Rev Ecol Syst 22: 477–503.

PLOS ONE | DOI:10.1371/journal.pone.0131082 August 5, 2015

21 / 21

Climate Tolerances and Habitat Requirements Jointly Shape the Elevational Distribution of the American Pika (Ochotona princeps), with Implications for Climate Change Effects.

Some of the most compelling examples of ecological responses to climate change are elevational range shifts of individual species, which have been obs...
NAN Sizes 0 Downloads 4 Views